demobook

Higgsfield: Automated workflow building with AI Assistant

Demo summary

The creator uses the built-in AI assistant to automatically generate a complex 20-node image testing pipeline from a single text instruction.

Step-by-step

  1. Open the built-in AI assistant within the canvas
  2. Enter a text instruction describing the desired workflow (e.g., 'takes my image prompt, optimizes it with an LLM, and tests it with every AI image gen model')
  3. Wait for the agent to build the nodes and instructions on the canvas
  4. Input your image prompt into the designated prompt node
  5. Click to generate images across all connected models simultaneously

Tips

  • Use the AI assistant to reorganize convoluted node layouts to make them cleaner
  • Provide a system prompt to the LLM node (like Claude) to ensure the prompt optimization works exactly as desired

Highlights

The agent built the entire canvas including instructions on how to use it in seconds.

All demos from “I Tested Higgsfield's New Canvas so You Don't Have to...

  1. 0:340:35Navigating the Higgsfield Canvas interfaceThe creator walks through the Higgsfield Canvas UI, demonstrating how to add nodes for image, video, and voice generation from the sidebar.HiggsfieldAI Video Generator
  2. 1:091:06Generate image prompts using LLM nodesThe demo shows connecting a headshot asset to a Claude Sonnet LLM node to generate an optimized prompt for a scuba diver image.HiggsfieldAI Image Generator
  3. 2:561:18Multi-model image generation in HiggsfieldThe user demonstrates connecting a single prompt to multiple image nodes (Nano Banana Pro, GPT Image 2, and Nano Banana 2) to generate different variations simultaneously.HiggsfieldAI Image Generator
  4. 4:140:38Convert image to video with node connectionsThe workflow is extended by connecting a generated image and an LLM-generated video prompt into a video generation node.HiggsfieldImage to Video
  5. 5:320:30Automated workflow building with AI AssistantCurrentThe creator uses the built-in AI assistant to automatically generate a complex 20-node image testing pipeline from a single text instruction.HiggsfieldAI Image Generator
  6. Watch “I Tested Higgsfield's New Canvas so You Don't Have to...” →

AI Image Generator

  1. 1:121:18Build a multi-model image generation workflow in Higgsfield CanvasThe user demonstrates how to add a prompt node and connect it to an image generator node, adjusting settings like aspect ratio and selecting models such as GPT image 2.RandomAI
  2. 2:301:37Benchmarking multiple AI models simultaneouslyThe creator shows how to connect a single prompt to three different image models (GPT image 2, Nano Banana 2, and Seedream 5.0 light) to compare their outputs side-by-side in one run.RandomAI
  3. 6:131:47Basic Image Generation with NodesA demonstration of creating a basic image generation workflow by connecting a prompt node to an image generation node using the GPT Image 2 model.DraftID
  4. 8:062:45Character Style Consistency with Prompt NodesThe video shows how to separate style and character prompts into different nodes to maintain a consistent 'miniature' look while swapping character subjects like an elf or a mech.DraftID
  5. 3:201:18Generate cinematic images with character and location tagsThe demonstration shows how to use '@' tags to reference saved characters and locations within a text prompt to generate consistent cinematic images in Higgsfield.Steven Wommack
  6. 11:132:02Compose and color grade a scene in HiggsfieldThe demo shows how to combine saved characters and locations into a single scene prompt, generate the image, and apply cinematic color grading and film grain effects.David Manning
  7. 2:091:54Build scenes in Higgsfield Cinema StudioThe user shows how to upload generated characters as references into Cinema Studio, select environments, and describe character interactions to create a static base scene.Money4Life
  8. 0:441:53Generate AI images with Higgsfield Canvas nodesThe user demonstrates how to connect a text prompt node and two image reference nodes to an image generator node in Higgsfield Canvas to create a specific cyberpunk character.Mhd Labs
  9. 0:321:14Generate an image using nodes in Higgsfield AIThe user demonstrates how to create a prompt node, connect it to an AI image generation node, and run the pipeline to generate an image of a cat within the Higgsfield AI canvas.Artturi Jalli
  10. 2:340:46Compare multiple AI image models on one canvasThe creator shows how to connect a single prompt to multiple different image generation nodes (like Cream 5.0 and Gro Imagine) to compare results side-by-side.Artturi Jalli
  11. 1:091:06Generate image prompts using LLM nodesThe demo shows connecting a headshot asset to a Claude Sonnet LLM node to generate an optimized prompt for a scuba diver image.Joseph Martin
  12. 2:561:18Multi-model image generation in HiggsfieldThe user demonstrates connecting a single prompt to multiple image nodes (Nano Banana Pro, GPT Image 2, and Nano Banana 2) to generate different variations simultaneously.Joseph Martin